Create igraph object from #asa2013 and #asa13

#This script creates a network of Twitter mentions and retweets
#It collects all tweets from the 2013 American Sociological Association with the hashtag #asa2013 or #asa13.
#Tweets without any relationship to another user of the hashtag are discarded.
#Twitter users mentioned who, if never they used the hashtags themselves, are also discarded

#The data was collected by Alex Hanna (https://twitter.com/alexhanna)
#I have reuploaded the data to pastein, however the original data can be found at:
#https://docs.google.com/spreadsheet/ccc?key=0Avriupr1jL4kdGxCOWlTZ3NWN28wTFFWcFRBV0l2UlE#gid=0

#The code below was written by Benjamin Lind (https://twitter.com/benjaminlind)
#Creative Commons [CC] license: BY-NC-SA 3.0


#This script may be loaded remotely using the following commands:
#library(RCurl)
#dl.data<-getURL("http://pastebin.com/raw.php?i=dY4VMMiF", ssl.verifypeer=FALSE)
#dl.data<-textConnection(dl.data)
#source(dl.data)
#rm(dl.data)

library(RCurl)
library(igraph)
library(compiler)

su<-function(x)
  return(sort(unique(x)))
su<-cmpfun(su)

read.sstab<-function(theurl, ...){ #Function to read in a data table from online
  #_theurl_ refers to the location of the data
  #_..._ are parameters passed onto read.table
  require(RCurl)
  outtab<-getURL(theurl, ssl.verifypeer=FALSE)
  outtab<-textConnection(outtab)
  outtab<-read.table(outtab, sep="\t", ...)
  return(outtab)
  }

dat<-read.sstab("http://pastebin.com/raw.php?i=CXjvtYZt", header=TRUE, as.is=TRUE, stringsAsFactors=FALSE, strip.white=TRUE, comment.char="", quote="")

dat<-dat[-which(dat$user.screen_name==""),] #Cleaning up
tweeters<-su(dat$user.screen_name) #Get a list of the user names

tweeters.mentioned<-lapply(tweeters, grep, x=dat$text) #Get a list of mentioned for each tweeter
names(tweeters.mentioned)<-tweeters #Name the list

cols.interest<-c("user.screen_name", "created_at", "id", "text", "rt.id", "rt.created_at") #The columns I'm interested in
ret.edge.info<-function(ind, d=dat)
  return(d[ind, cols.interest])
ret.edge.info<-cmpfun(ret.edge.info)

el.make<-function(sn, l=tweeters.mentioned){ #Generates the edgelist
  colnames.mat<-c("sender", "receiver", "time", "id", "text", "rt.id", "rt.time")
  if (length(l[[sn]])==0){ #Send out NA vector if person is never mentioned
    ret.vals<-matrix(NA, nrow=1, ncol=length(colnames.mat), dimnames=list("1", colnames.mat))
    return(ret.vals)
    }
  else{
    ret.vals<-lapply(l[[sn]], ret.edge.info)
    ret.vals<-do.call(rbind, ret.vals)
    ret.vals<-as.matrix(ret.vals)
    if (nrow(ret.vals)==1){ #Ensure that it returns a matrix, not a vector
      ret.vals<-c(sender=ret.vals[,"user.screen_name"], receiver=sn, ret.vals[c("created_at", "id", "text", "rt.id", "rt.created_at")])
       ret.vals<-matrix(ret.vals, nrow=1, ncol=length(ret.vals), dimnames=list(1, colnames.mat))
       }
    else{
      ret.vals<-cbind(sender=ret.vals[,"user.screen_name"], receiver=rep(sn, times=nrow(ret.vals)), ret.vals[, c("created_at", "id", "text", "rt.id", "rt.created_at")])
      ret.vals<-as.matrix(ret.vals, dimnames=list(1, colnames.mat))
      }
    return(ret.vals)
    }
  }; el.make<-cmpfun(el.make)

tweeter.el<-lapply(tweeters, el.make)
tweeter.el<-do.call(rbind, tweeter.el)
tweeter.el<-as.data.frame(tweeter.el)
tweeter.el$rt.id<-as.numeric(tweeter.el$rt.id) #Convert from character to numeric
tweeter.el$id<-as.numeric(tweeter.el$id) #Convert from character to numeric
tweet.date<-as.numeric(as.Date(tweeter.el$time, "%m/%d/%Y %H:%M:%S"))  #Numeric date
tweet.rt.date<-as.numeric(as.Date(tweeter.el$rt.time, "%m/%d/%Y %H:%M:%S")) #Numeric date
tweeter.el<-cbind(tweeter.el, time.num=tweet.date, rt.time.num=tweet.rt.date, is.rt=is.na(tweeter.el$rt.id)==FALSE) #Add new, numeric date vectors, along with RT logical vector
tweeter.el<-tweeter.el[-which(is.na(tweeter.el$sender)&is.na(tweeter.el$receiver)),] #Remove those without mentions

tweeter.ig<-graph.edgelist(as.matrix(tweeter.el[, c("sender", "receiver")])) #Turn to edgelist igraph
for (i in 3:ncol(tweeter.el)) #Add edge attributes
  tweeter.ig<-set.edge.attribute(tweeter.ig, name=colnames(tweeter.el)[i], value=tweeter.el[,i])
unmentioned<-tweeters[which(tweeters %in% V(tweeter.ig)$name==FALSE)] #Who tweeted, but wasn't mentioned?
tweeter.ig<-add.vertices(tweeter.ig, nv=length(unmentioned)) #Add isolates
V(tweeter.ig)$name[which(is.na(V(tweeter.ig)$name))]<-unmentioned
#Set the vertex attributes
dat.mini<-dat[which(duplicated(dat$user.screen_name)==FALSE), c("user.screen_name", "user.id", "user.name")]
V(tweeter.ig)$user.id<-dat.mini[match(V(tweeter.ig)$name, dat.mini[,"user.screen_name"]), "user.id"]
V(tweeter.ig)$user.name<-dat.mini[match(V(tweeter.ig)$name, dat.mini[,"user.screen_name"]), "user.name"]

rm(su, read.sstab, dat, tweeters.mentioned, ret.edge.info, cols.interest, el.make, tweet.date, tweet.rt.date, i, unmentioned, dat.mini, tweeter.el, tweeters) #Clean-up